Skip to content

Moseq2 Bucket Preparation

Sherry edited this page Jan 22, 2024 · 4 revisions

Sanding

1.) Use 80-grit sandpaper. Sand the floor firmly with consistent ~2" circular motions. Remove plastic debris as you go.

2.) The goal here is to completely sand over the original smooth surface. There should be no smooth or reflective areas. Any patches of smooth surfaces will cause IR reflectance and critically disturb your Kinect data.

3.) After the floor is completely sanded, turn the bucket on its side and sand the walls using the same motion from Step 1 – it’s only necessary to sand the lower 8” of the wall.

4.) Be very sure that the surface is uniformly sanded and that there are no reflective smooth spots. This is particularly important at the wall-floor junction, where the mouse will spend a lot of its time.

Sanding the bucket properly is essential for acquiring high-quality data. Here's an example of a mouse recorded in a bucket with little to no sanding:

As you can see, when the mouse approaches the wall, a major artifact (IR glare) appears. (If this is happening to you, try re-sanding your bucket)

Most of the time you will get some speckled noise along the wall, which is permissible:

If you do an exceptional job sanding, you can get nearly noiseless data:

Use this bucket comparison as a guide as you sand!

Note: The difference between "Good" and "Bad" sanding is subtle. The crucial difference between the two is that the "Good" sand is more uniformly rough and there are no small areas of smooth material. This is particularly noticeable in the top->down picture. It will take multiple "coats" of sanding to achieve "Good" quality.

Cleaning

We clean our buckets before a new mouse is put in the arena with the following solutions (in spray bottles), in this order, and wipe the arena dry with a paper towel after each is used:

  1. 10% bleach
  2. 1% Alconox
  3. 70% Ethanol

Use the ethanol conservatively, as it eats away at the spray paint over time. We have buckets that have been cleaned with ethanol for 2+ years and haven't noticed appreciable differences in data quality. However, it is conceivable that it could be necessary for the future to touchup the bucket up with another coat of spray paint.

Clone this wiki locally